Modify the gen_pool allocator (lib/genalloc.c) to utilize a bitmap scheme
instead of the buddy scheme. The purpose of this change is to eliminate
the touching of the actual memory being allocated.
Since the change modifies the interface, a change to the uncached allocator
(arch/ia64/kernel/uncached.c) is also required.
Both Andrey Volkov and Jes Sorenson have expressed a desire that the
gen_pool allocator not write to the memory being managed. See the
following:
http://marc.theaimsgroup.com/?l=linux-kernel&m=113518602713125&w=2http://marc.theaimsgroup.com/?l=linux-kernel&m=113533568827916&w=2
Signed-off-by: Dean Nelson <dcn@sgi.com>
Cc: Andrey Volkov <avolkov@varma-el.com>
Acked-by: Jes Sorensen <jes@trained-monkey.org>
Cc: "Luck, Tony" <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
There is an SN bug in sn_hwperf.c that affects systems with 1024n or 1024p.
The bug manifests itself 2 ways: IO interrupts are not always
targeted to the nearest node, and 2) the "cat /proc/sgi_sn/sn_topology"
commands fails with "cannot allocate memory".
The code is using the wrong macros for validating node numbers.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
MPI programs using certain debug options have a long
startup time. This was traced to a "vmalloc/vfree" in
the code that reads /proc/sgi_sn/sn_topology. On large
systems, vfree requires an IPI to all cpus to do TLB
purging.
Replace the vmalloc/vfree with kmalloc/kfree. Although
the size of the structure being allocated is unknown, it
will not not exceed 96 bytes.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Fix a bug that causes discovery of the nearest node/cpu to
a TIO (IO node) to fail.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
[IA64] ioremap() should prefer WB over UC
[IA64] Add __mca_table to the DISCARD list in gate.lds
[IA64] Move __mca_table out of the __init section
[IA64] simplify some condition checks in iosapic_check_gsi_range
[IA64] correct some messages and fixes some minor things
[IA64-SGI] fix for-loop in sn_hwperf_geoid_to_cnode()
[IA64-SGI] sn_hwperf use of num_online_cpus()
[IA64] optimize flush_tlb_range on large numa box
[IA64] lazy_mmu_prot_update needs to be aware of huge pages
Tweak the proc setup code so things work OK with const
proc_dir_entry.proc_fops.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Fix a for-loop in sn_hwperf_geoid_to_cnode(). It needs to loop over
num_cnodes to ensure it can still process TIO nodes in addition to
compute nodes on systems with many nodes. Interim fix until better
support for many (>265) nodes is complete.
Signed-off-by: Dean Roe <roe@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Eliminate an unnecessary -- and flawed -- use of the expensive
num_online_cpus().
Signed-off-by: John Hawkes <hawkes@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
This stuff is all in the generic ia64 kernel, and the new initcall error
reporting complains about them.
Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
The original ia64 udelay() was simple, but flawed for platforms without
synchronized ITCs: a preemption and migration to another CPU during the
while-loop likely resulted in too-early termination or very, very
lengthy looping.
The first fix (now in 2.6.15) broke the delay loop into smaller,
non-preemptible chunks, reenabling preemption between the chunks. This
fix is flawed in that the total udelay is computed to be the sum of just
the non-premptible while-loop pieces, i.e., not counting the time spent
in the interim preemptible periods. If an interrupt or a migration
occurs during one of these interim periods, then that time is invisible
and only serves to lengthen the effective udelay().
This new fix backs out the current flawed fix and returns to a simple
udelay(), fully preemptible and interruptible. It implements two simple
alternative udelay() routines: one a default generic version that uses
ia64_get_itc(), and the other an sn-specific version that uses that
platform's RTC.
Signed-off-by: John Hawkes <hawkes@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
General SN2 code cleanup:
- Do not initialize global variables to zero
- Use kzalloc instead of kmalloc+memset
- Check kmalloc return values
- Do not obfuscate spin lock calls
- Remove some unused code
- Various formatting cleanups
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Prevent SN2 specific code to be executed on non SN2 platforms when
running a generic kernel.
Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
On SN2, MMIO writes which are issued from separate processors are not
guaranteed to arrive in any particular order at the IO hardware. When
performing such writes from the kernel this is not a problem, as a
kernel thread will not migrate to another CPU during execution, and
mmiowb() calls can guarantee write ordering when control of the IO
resource is allowed to move between threads.
However, when MMIO writes can be performed from user space (e.g. DRM)
there are no such guarantees and mechanisms, as the process may
context-switch at any time, and may migrate to a different CPU as part
of the switch. For such programs/hardware to operate correctly, it is
required that the MMIO writes from the old CPU be accepted by the IO
hardware before subsequent writes from the new CPU can be issued.
The following patch implements this behavior on SN2 by waiting for a
Shub register to indicate that these writes have been accepted. This
is placed in the context switch-in path, and only performs the wait
when the newly scheduled task changes CPUs.
Signed-off-by: Prarit Bhargava <prarit@sgi.com>
Signed-off-by: Brent Casavant <bcasavan@sgi.com>
This patch finishes support for SHUB2 (the new chipset). Most of the
changes are performance related. A few changes are workarounds for
"interesting" chipset features.
Some temporary debugging code has also been deleted.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
arch/ia64/sn/Makefile sets CPPFLAGS, expecting that setting to
propogate to all the subdirectories. For a normal build with its
recursive descent it does work, but doing a selective build like
'make arch/ia64/sn/kernel/io_init.i' does not do a recursive descent,
it goes directly to arch/ia64/sn/kernel/Makefile so the flags do not
get set.
To support selective builds, set the flags in all the subordinate Makefiles.
Signed-off-by: Keith Owens <kaos@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
I see why the problem exists only on SN. SN uses a different hardware
mechanism to purge TLB entries across nodes.
It looks like there is a bug in the SN TLB flushing code. During context switch,
kernel threads inherit the mm of the task that was previously running on the
cpu. This confuses the code in sn2_global_tlb_purge().
The result is a missed TLB purge for the task that owns the "borrowed" mm.
(I hit the problem running heavy stress where kswapd was purging code pages of
a user task that woke kswapd. The user task took a SIGILL fault trying to
execute code in the page that had been ripped out from underneath it).
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
The patch that added support for a new platform chipset (shub2) broke
PTC deadlock recovery on older versions of the chipset. (PTCs are the
SN platform-specific method for doing a global TLB purge). This
patch fixes deadlock recovery so that it works on both the old & new
chipsets.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
The kernel's use of the for_each_*cpu(i) macros has allowed for sparse CPU
numbering. When I hacked the kernel to test sparse cpu_present_map[] and
cpu_possible_map[] cpumasks, I discovered one remaining spot, in
sn_hwperf_ioctl() during sn initialization, that needs to be fixed.
Signed-off-by: John Hawkes <hawkes@sgi.com>
Signed-off-by: Dean Roe <roe@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Patch to prevent sn2_ptc_init code from attempting to load on non-sn2 systems
when sn2_smp.c is built-in to generic kernel.
Signed-off-by: Prarit Bhargava <prarit@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
flush_tlb_all() can be a scaling issue on large SGI Altix systems
since it uses the global call_lock and always executes on all cpus.
When a process enters flush_tlb_range() to purge TLBs for another
process, it is possible to avoid flush_tlb_all() and instead allow
sn2_global_tlb_purge() to purge TLBs only where necessary.
This patch modifies flush_tlb_range() so that this case can be handled
by platform TLB purge functions and updates ia64_global_tlb_purge()
accordingly. sn2_global_tlb_purge() now calculates the region register
value from the mm argument introduced with this patch.
Signed-off-by: Dean Roe <roe@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Some of the SN code & #defines related to compact nodes & IO discovery
have gotten stale over the years. This patch attempts to clean them up.
Some of the various SN MAX_xxx #defines were also unclear & misused.
The primary changes are:
- use MAX_NUMNODES. This is the generic linux #define for the number
of nodes that are known to the generic kernel. Arrays & loops
for constructs that are 1:1 with linux-defined nodes should
use the linux #define - not an SN equivalent.
- use MAX_COMPACT_NODES for MAX_NUMNODES + NUM_TIOS. This is the
number of nodes in the SSI system. Compact nodes are a hack to
get around the IA64 architectural limit of 256 nodes. Large SGI
systems have more than 256 nodes. When we upgrade to ACPI3.0,
I _hope_ that all nodes will be real nodes that are known to
the generic kernel. That will allow us to delete the notion
of "compact nodes".
- add MAX_NUMALINK_NODES for the total number of nodes that
are in the numalink domain - all partitions.
- simplified (understandable) scan_for_ionodes()
- small amount of cleanup related to cnodes
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Clean up of SGI SN partitioning related code.
The SN_SAL_GET_SN_INFO SAL call returns the partition ID, making
the SN_SAL_SYSCTL_PARTITION_GET SAL call redundant. Remove sn_partid
and use sn_partition_id.
Signed-off-by: Russ Anderson (rja@sgi.com)
Signed-off-by: Tony Luck <tony.luck@intel.com>
Add a new exported function for determining the nearest node
with CPUs for I/O nodes and fix a bug where the hwperf dynamic
misc device was being registered before misc_init().
Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Bugfix to export PCI topology information in /proc/sgi_sn/sn_topology.
Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Removed IA64 architecture specific users of asm/segment.h
The removal of asm-ia64/segment.h itself can wait until all
of the kernel source has been purged of references.
Signed-off-by: Kumar Gala <kumar.gala@freescale.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Shub2 provides a much improved mechanism for issuing internode
TLB purges. Add code to support the newer mechanism. There is also
some debug code (disabled) that is useful for testing.
Collect statistics on the number, type & duration of TLB purges.
This data will be useful for making future improvements in the algorithms.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Disable some shub1-specific code when running on systems with shub2.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
This patch is a sparse compile cleanup of shub_mmr.h using both the defconfig
and the sn2_defconfig config files.
The issue with this file was the missing usage of __IA64_UL_CONST wrapper.
This wrapper is defined in include/asm-ia64/types.h and wraps a long
constant definition with UL or with nothing depending on its usage in the
kernel. The missing wrapper caused many sparse compile errors like
warning: constant 0x0x0000000010000380 so big it is long
Signed-off-by: Prarit Bhargava <prarit@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
This patch against ia64-test-2.6.12 is needed for forthcoming
Altix chipsets. It renames geoid_any_t to geoid_common_t and
splits the 8bit 'slab' field into two 4bit fields for 'slab'
and 'slot'. Similar changes in the Altix SAL will retain backward
compatibility for old kernels.
Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
please accept this patch to the Altix SN platform topology export
interface to support new chipsets and to export PCI topology.
This follows on top of Jack Steiner's patch dated March 1st
("New chipset support for SN platform").
Signed-off-by: Mark Goodwin <markgw@sgi.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Jörn Engel
Tony Luck
Dean Nelson
Jack Steiner
Linus Torvalds
Andrew Morton
Dean Roe
hawkes@sgi.com
Bjorn Helgaas
Jes Sorensen
Brent Casavant
Keith Owens
Prarit Bhargava
Len Brown
Russ Anderson
Mark Goodwin
Kumar Gala
Linus Torvalds